Originally Posted by sspears
Our window patterns are more accurate than virtually every HW pattern generator because of our patented dither that allows us to hit the correct percent. e.g. Most 50% patterns use code value 126 in 8-bit. The correct value is 125.5, which we can achieve by using dither. In 10-bit, this is even worse because some take the incorrect 8-bit value of 126 * 4, which results in 504. The correct value in 10-bit is actually 502. (940-64 = 876. 876 / 2 + 64 = 502) In 8-bit this is 235-16 = 219. 219 / 2 + 16 = 125.5. Sorry for the math, just wanted to point out subtle differences and the type of detail that we pay attention to.
Hi Stacey, as this thread talk for S&M2, the patterns are 8bit, but as you used more sophisticated method to create them (using still frames helps also), even Murideo is recommending to use your patterns instead of using Murideo's gradation patterns.
If you measure in SDR a 10-bit pattern with digital code 502 vs. 504, the meter repeatability and panel instability will not able to get you any difference. (the calibration software will do the target calculation in 126, so in 504 10-bit, not for 506 which is the correct 10-bit)
The point is that any calibration software we know (LightSpace/HCFR/CalMAN/ChromaPure) do all the color engine calculations based to 8-bit RGB triplets, so there currently no benefit when you will use the most accurate 10-bit patterns, as it will deviate more from the 8bit triplets the software will expect from you to display.
The advantage of having perfect 10-bit patterns will be useful to all the other areas (ramps/gradation for example) but not to solid patch colors for measurements area, it will have benefit when calibration software will be able to generate/calculate true 10-bit values, useful for HDR calibrations. But in HDR mode the panels are so unstable with poor repeatability, so it will not be any real advantage in measurements.
All calibration enthusiast users need to have at least S&M2 + S&M3 disk to their collection for sure. (before its sold out)
Originally Posted by sspears
There is another issue and that is some displays multi-point grayscale don't align with standard spacing. We do not deal with this case. We build patterns to standards and we don't work around non-standard layouts. Again, this is where a HW pattern gen driven by software can deal with those cases. Some discs have created custom patch sets to deal with this. e..g Ryan has some patch sets for older LGs that had some odd spacing. We are not going to to do that.
This is happening because LG calibration controls are adjusting values in gamma (SDR) space and then its added PQ math, so the 20-Point controls in HDR manual cal, are dynamically adjusted based to the user adjustments, for that reason its impossible to create a specific patch set for that display which will work for all users while its so easy banding to introduced when you will try large adjustments or weird compo of adjustments.
Ryan's HDR patchsets can help but not resolve completely the problem.
The problem will be resolved when LGE, as the OLED natively has linear 0-1 response, but the TCON module is programmed to make them 2.2 gamma based....If LG will change that and provide additional data for the panel to have a PQ response also, then there will be no problem with alignment of the HDR patterns then. (as the controls will be based to PQ values)
As this will require a lot of changes completely to the panel calibration and factory manufacturing process changes, it will need changes of the whole engine/video processing of main board etc... its not so easy for this to happen, as only OLED are linear, the LCD's are not.